Energy meter with fast cycling load

[Skogsgurra]

I have on my bench a kWh meter that is supposed to have a systematic (all meters of that type seem to have the problem) error.

Calibration works out fine, but when you cycle the load with a few seconds on and a few seconds off, the meter registers more energy than is actually used by the load.

The reason this deviation was detected is that a guy with a (I think) badly designed thermostat in an around 2 kW application thought he paid too much for electricity. Utility did a check and found that the meter actually registered too much energy. Not when testing with steady load, but when checking with parallel meter from other manufacturer.

I have some theories about this. But feel genuinly unsure. My first thought is that some accumulating or filtering software is asymmetrical so that 'attack' is faster than 'decay' - if I am allowed to use musical language.

Anyone has any experience with this? Heard about a similar thing? I am open for all possibilities - the weirder the better (well, within limits).

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[waross]

Possibly a slow sample rate? Some interaction between the sample rate and the load cycle time?
Did you hear about the retired railway man with the sleep disorder? He would wake at random times for a few hours and go back to sleep to wake again at a truly random time. When he awoke he would go down to the tracks and wait for the next train to pass. He would note the direction of the train and return home to sleep again. He saw three times as many east bound trains as west bound trains. How could this be? Actually there were the same number of trains each way, but the eastbound train passed on the hour and the west bound train passed 15 minutes later.
He had a 15 minute window of opportunity to see a west bound train and a 45 minute window of opportunity to see an east bound train.
Random sampling produced a 300% error.
Is it possible to have a similar interaction between the sampling rate and the load cycling rate?

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Skogsgurra]

Have been thinking about that. The meter manufacturer says he samples 1000 times/second. And I really hope that this is for all six (three voltage and three current) channels. That would be 20 samples/period (50 Hz) and it does not seem possible to have an error in the 15 percent range with such a tight sampling. Well, not exactly tight, but good enough for a meter, I think.

If the sampling is spread over all six channels, then anything could happen, 3.33 Sa/period and switching load on/of in seconds could produce some interesting interference patterns. But ought to equal out in the long run. Also, this is a well-known manufacturer and the specs (checked by independent lab) are good. So, I think we have 20 Sa/period on each channel.

I am setting up a test with triacs for fast cycling and a bank of 9 times 300 W lamps. It is getting bright (and hot) when we turn them on.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[waross]

Don't forget to check your own kWHr meter and bill the customer for the power consumed during testing. LOL

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Skogsgurra]

Good tip! I usually lose more than I gain. Billing this energy, which is heating my house, I will perhaps come out even.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[itsmoked]

Could be they are getting spikes coupling in. Essentially blunders in data. Is the customer using random SSRs?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

I shall have a closer look at the current recorded. Stand by.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Skogsgurra]

Looks like turn-on is completely random. Turn-off is, of course dictated by zero crossing. The currents (channels) are 1, 3 and 5 (resistor colour code). Upper recording shows turn-on and it looks like thyristors. Lower recording confirms that. The load is obviously connected L2 - L3 with a small load connected L1-N. A small switch-mode power supply seems to be connected from L2 to N (can be seen in lower picture after cursor CIII.

The length of this packet is 8 seconds (Delta C3-C2) and the off time is 7 seconds (Delta C2-C1).

Not much distortion.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[itsmoked]

This is the 'customers' load?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[electricpete]

I'm a little confused what we're looking at as well. We are evaluating a test signal used to check a meter?

But interesting to see the waveforms do not go to zero during the off-time. Is there maybe measured energy accumulating during the off-time?

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Skogsgurra]

Thanks for your comments!

First, yes, this is what the customer's load looks like. It is an ordinary one-family house at around ten o'clock in the morning. There are very few other loads except the heater that is connected between L2 and L3. When the heater is on, current flows from L2 to L3, so channel 3 and 5 shall be 180 degrees apart.

Then, when the heavy load is off, there is something with o little crest. Probably a switchmode power supply (TV set or computer) in channel 3.

Channel 1 was off in the upper picture, but is activated in lower picture. Looks like a modest linear load. It was mentioned that a small heater with a fan was on in addition to the heater.

There seems to be a DC offset in channel 5 lower picture. I hadn't noticed that before. Me to wonder from where that comes. Shall look for it in my following recordings.


I am now trying to load the meter with a set of incandescent lamps controlled by a three-phase thyristor bridge that is, in its turn. controlled by a function generator so I can have rapid cycling and really nasty waveforms. Trying to provoke bad behaviour.

My first test will be to check if there really is a fast 'attack' and a slow 'release' when power is cycled. If there is a linear filter involved, I think that cycling should have no effect on the accumulated energy. If the filter behaves differently when switching power on and switching it off, I think that there could be a net error.


Thanks again. Six eyes are a lot better than two.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[itsmoked]

What is the actual measurement device in the meter? Is it a uC or an application specific device that does-it-all?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

We are not supposed to know. There's a seal on the meter and I do not want to break it.
These meters are made in millions, so I guess that there's a lot of application specific stuff inside.

Good thing: Got my thyristors working, really nasty waveforms and cycling at several Hz. My lab has turned to a disco.

Now I have a 'slight' Excel problem; how to measure distance between the metering pulses. Each pulse is around 60 rows in the sheet. I have done edge detection and have a resulting '1' for each positive edge. Now looking for a way to measure distance (seconds) between the edges. Never done that before. Guess I can use row number when I find a cell with '1' in it and subtract row number of previous cell with a '1'. Not what I usually do.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[electricpete]

Sounds like the excel part would be easy - let us know here or in excel forum if you want ideas on that.

I don't work with meters, but the whole thing is mysterious to me. I would have thought a kw-hr meter would simply accumulate continuously using simple stepwise integration of instantenous power (V*I) to get energy W:

Wk+1 = Wk + (Ik+1 * Vk+1)*DeltaT .... where DeltaT is time between samples.

Apparently it is different than that because there is something called a "metering pulse". What is the purpose of a metering pulse? ... When pulse occur the power is computed (based on a half cycle or cycle interval), and used as representative of the entire interval between pulses?

In either case, I'm not sure what would be the purpose of a filter. I guess I can see the 2nd algorithm (metering pulse) where we look at a short shapshot 1-cycle or so would be susceptible to small deviations in power frequency... if the samples do not occur at exactly submultiples of power frequency period than a snapshot 10 or 20 samples would have some error based on looking at an interval slightly different than a full or half cycle.... and it power frequency was consistently different in same direction that error would not be random. Might need something more complicated to deal with that which I suppose could include a filter. But nothing here to do with cycling loads.

Do you know yet the approximate time between pulses?

What is the magnitude of the error that is seen?

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Skogsgurra]

The metering pulse is used to transmit energy data to a remote meter reading system like Turtle or some other PLC (Power Line Communication) system. Or to a radio communicatiob channel.

Metering pulses are short (20 - 40 ms) pulses that corresond to a certain amount of energy. In this case 1000 pulses correspond to 1 kWh. Distance between pulses can be seen from attached recording. It is from the same measurement as the ones shown before. But this is taken over longer time. The metering pulses are on channel 6 (blue). Currents in the three phases are shown in channels 1, 3 and 5 as before.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[electricpete]

Thanks for explaining that.... makes sense now, although I still don't see the need for any filtering. Another observation I'm sure you noticed, it appears there is a one or two second delay in generating the pulses. Not a big deal for the application but just complicates your analysis task slightly.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Skogsgurra]

That's the problem. I don't see any need for it either. And still, it looks like there is something that influences the pulse output. And, if the load is cycled, it looks like it affects the accumulated energy. That is why I am doing a test with faster cycling.

The viewer program is available at

http://www.gke.org/test/

and I will put the original file on

http://www.gke.org/pub

for download. Data files with faster cycling will also be uploaded soon.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[electricpete]

The following user-defined function may help. Can find the row number of a 1 anywhere within specified range.

= FindOne(3,A1:A5,False) would find the 3rd occurence of the number 1 from the bottom in range A1:A5

Function FindOne(Occurence As Integer, myrange As Range, SearchDown As Boolean) As Integer
' Finds the Occurence_th occurence of the number 1 within range,
' (For example if Occurence = 1, it finds the first occurence, if Occurence=2, if finds 2nd)
' it searches downwards from top if SearchDown is true
' it searches upwards from bototm if SearchDown is false


' Note you can use the built-in match function if you simply want to search down for first occurence
' Assumes the input range is a single-column range... no error checking is done

Dim rowcounter As Integer ' count rows of range
Dim foundcounter As Integer ' count number of 1's found
Dim thisrow As Integer ' identifies row number for the loop
Dim thiscell As Range ' identifies cell for the loop

    FindOne = 0   ' Error flag - if function returns 0 there are not enough 1's in the range
    foundcounter = 0 ' Initialize

    For rowcounter = 0 To myrange.Rows.Count - 1
        ' Set thisrow according to User Preference in SearchDown
        If SearchDown = True Then
            thisrow = rowcounter
        Else
            thisrow = myrange.Rows.Count - rowcounter
        End If

        ' Identify the cell of interest
        Set thiscell = myrange.Offset(thisrow, 0).Resize(1, 1)

        ' Increment foundcounter if cell contains 1
        If thiscell.Value = 1 Then
            foundcounter = foundcounter + 1
        End If

        ' If foundcounter matches target, output results and terminate
        If foundcounter = Occurence Then
            FindOne = thiscell.Row
            Exit For
        End If
    Next rowcounter
End Function

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Skogsgurra]

Thanks a lot, Pete!

I had a feeling you would jump in and help. Much appreciated. Only, I have never done user defined functions. Something new to learn!

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[electricpete]

Glad if I can help to return the favor for ideas about checking motor rotation among other things.

Here's how to use the function (at least in my excel 2003)

In the workbook of interest:

Select Tools / Macro / VisualBasicEditor

Select Insert / Module

That should open a subwindow which says "General" near the top.
Paste the entire code listed in my post above into that subwindow

Perhaps double check on left side that the selected module (module 1 probably) is in a folder named modules which appears underneath something called 'VBA Project ("MyFile.xls")'.... where MyFile.xls is the name of your spreadsheet. This just verifies you are working in the right place, rather than inserting into your personal.xls or some other vba window.

Select File/CloseAndReturnToMicrosoftExcel


=====================================
Eng-tips forums: The best place on the web for engineering discussions.